Five state-backed hackers caught using ChatGPT for crimes, promptly shown the door
Ourtake
Turns out the hacking assistant also needed a translator, a debugger and a shutdown notice.
What happened
OpenAI and Microsoft Threat Intelligence shut down five accounts linked to state-affiliated threat actors from China, Iran, North Korea and Russia. The actors used OpenAI models for research, translation, code debugging and creating malicious scripts or phishing content. OpenAI said the activity was consistent with prior red-team findings that GPT-4 offers only limited, incremental capabilities for malicious cybersecurity tasks.